why do people polish face plates of coolers?

[bob4933]

Speaking of the mating surface, not exposed parts!

Never understood this... its not something you can see... I see it done a lot though and don't really get the point.

 

[jaymz9350]

The main reason it to flatten the surface for better contact. Many coolers and IHS's are not flat so the two pieces do not make complete contact. The polishing (making it shiny) doesn't really make a big difference as far as I know but it looks cool lol.

 

[ehume]

The finer the shine, the fewer grooves and the better the surfaces mate.

 

[bob4933]

Those "grooves" help increase surface area, which will have better cooling though. It's pretty well documented that polishing a surface reduces heat transfer.

 

[\dev\goat]

http://www.overclockers.com/forums/showthread.php?t=716889

There's the thread from when I lapped my old corsair H80 and 2600k.

The evidence is entirely anecdotal, because I didn't think far enough ahead to capture any "before" data. But I did see a fairly remarkable improvement in my temps.

 

[manu2b]

Those "grooves" help increase surface area, which will have better cooling though. It's pretty well documented that polishing a surface reduces heat transfer.

True if the TIM have a 100% heat transfer capacity, which is never the case, far from that.

So you need the heatsink to be as much as possible in contact with the IHS to reduce TIM thickness/quantity.

 

[m0r7if3r]

Those "grooves" help increase surface area, which will have better cooling though. It's pretty well documented that polishing a surface reduces heat transfer.

Not a surface which will be in contact with multiple media. The TIM does not transfer heat anywhere near as well as metal-to-metal contact (but it transfers heat better than air). By increasing metal-to-metal contact, you maximize heat transfer out of the CPU and into the HS.

Alternatively, think of it this way: if it were possible to get the two surfaces perfectly flat (it's not), there would be no need for TIM. At THAT point, it might begin to make sense to attempt to increase surface area on the IHS and HS, but only if you could do it while maintaining perfect or near-perfect interface between the materials (which is impossible).

 

[bob4933]

Hmm, I am probably being too analytical then. Where is the weak link then? The whole point of TIM is not to improve metal-to-metal conductivity, its to remove air, which is an insulator.

I am a mechanical engineer and I work at a nuclear power plant, so all of this heat transfer business is pretty near and dear to my heart. On our steam generators, we actually do a scrubbing process to increase the amounts of surface roughness because when the plates are trued up they are too "finely finished" (yes, they are about as flat as you can get, laser defined leveling process, CAD controlled, pretty damn accurate haha), and polishing the surface would reduce the performance of the steam generators (more than you would expect!).


Devgoat -> I appreciate that bit, however, did you use the same TIM? Did you apply the same amount in the same way? Same ambient temps? Thats just not a proper "test" to define this honestly.

Now if there are actual untrue grooves and markings/warpings in the plates, then absolutely this is necessary, I've just never seen a plate or CPU that seemed to "need" polishing. If thats the case, then it makes a lot more sense to me. Getting an even surface is far more important than any "polishing". I'm just seeing a lot of "polished" for polished sakes out there, and it just seems really strange to me. Simple polishing will not true up anything, and sometimes will make things worse if you use a silicone based compound. Thats more of what Im speaking of.

 

[\dev\goat]

Devgoat -> I appreciate that bit, however, did you use the same TIM? Did you apply the same amount in the same way? Same ambient temps? Thats just not a proper "test" to define this honestly.

Right, it was purely anecdotal

I did use the same TIM, and used the same pattern (small blob, let the mounting pressure do the spreading).

After mounting and removing it, the spread pattern of the TIM was entirely different after lapping both surfaces, though. Before lapping, it was spotty and patchy. Afterwards, it would make a neat little evenly-spread square.

I think much of the gains I saw came from lapping the heatsink, though, as it was visibly concave and had marks all over it from the milling. Lapping the cpu probably just gave a small gain, from reducing the thickness of the heat spreader.

 

[bob4933]

Right, it was purely anecdotal

I did use the same TIM, and used the same pattern (small blob, let the mounting pressure do the spreading).

After mounting and removing it, the spread pattern of the TIM was entirely different after lapping both surfaces, though. Before lapping, it was spotty and patchy. Afterwards, it would make a neat little evenly-spread square.

I think much of the gains I saw came from lapping the heatsink, though, as it was visibly concave and had marks all over it from the milling. Lapping the cpu probably just gave a small gain, from reducing the thickness of the heat spreader.

I can see it noticeably helping if the surfaces werent mating properly, and from that end makes A HUGE amount of sense to me of course. I just see people using brasso, and I just kinda...

 

[Pierre3400]

Those "grooves" help increase surface area, which will have better cooling though. It's pretty well documented that polishing a surface reduces heat transfer.

This is indeed 100% spot on correct.

But the fact is, what you are talking about is a surface that is free to be cooled.


Aircooled cylinder, found many places, e.g an air compressor. Has fins to help it cool, but in the case of a CPU and a heat sink, we have a contact patch between 2 surfaces that have to transfer heat and not get rid of heat.

The better the transfer of the heat to the heatsink, the better the cooling can be. The surface you are looking for, its the surface on the fins of the heatsink.

The polish on the CPU and/or heatsink comes not from polishing, but using sand paper of such fine grit it shines.

 

[bob4933]

No, I literally mean, people take brasso or some other polishing compound to their coolers. Never understood where that started, or why people think its a good idea. I am thinking there is some language barrier to what you're trying to get at pierre, sorry . If you're saying Im misguided about what Im thinking of, no I am not. I literally mean, that there is a nominal amount of surface roughness that actually raises heat transfer capabilities of a material, with or without a thermal compound. Polishing metals to a mirror like shine is detrimental to heat transfer when all other things are considered equal (i.e. the surfaces are trued up).



But...

For example... this is just silly. Looks "cool" but sheesh...

 

[MattNo5ss]

Using very fine sandpaper to get the surface so smooth that it looks like a mirror allows you to use less TIM since the mating surfaces will be closer together. Any potential performance increases by more surface area of the mating surfaces becomes irrelevant because of the decrease in performance caused by having to use more TIM in that case.

Using some type of polish just to make the surface shine is pointless if it's not actually making the surfaces smooth.

 

[Tyerker]

No, I literally mean, people take brasso or some other polishing compound to their coolers. Never understood where that started, or why people think its a good idea. I am thinking there is some language barrier to what you're trying to get at pierre, sorry . If you're saying Im misguided about what Im thinking of, no I am not. I literally mean, that there is a nominal amount of surface roughness that actually raises heat transfer capabilities of a material, with or without a thermal compound. Polishing metals to a mirror like shine is detrimental to heat transfer when all other things are considered equal (i.e. the surfaces are trued up).



But...

For example... this is just silly. Looks "cool" but sheesh...

Brasso is more or less super-fine liquid sandpaper. So I get if you get to like 2000 grit sandpaper, then use some Brasso, maybe that will help a little bit. Brasso (and most other polishing compounds) does take off metal, though.

 

[bob4933]

Brasso is more or less super-fine liquid sandpaper. So I get if you get to like 2000 grit sandpaper, then use some Brasso, maybe that will help a little bit. Brasso (and most other polishing compounds) does take off metal, though.

All polishing compounds take off metal, thats simply how polishing works. That said, you can't "true up" a surface with a polish, just not feasible, certainly not by hand either.

Its also the fact that it leaves a chemical residue that impedes your heat transfer capabilities that strikes me as odd. Not just brasso, but tons of polishing compounds use silicone.

 

[Robert17]

There is another angle (pun intended) to it from my experience. I recently moved my H60 from one CPU to another. The plate on the H60 would not release all the TIM after many wipes with a paper towel, soft cloth, even with isopropyl alcohol, surface roughness being what it was. So I applied two different grits of rouge for about 10 minutes of polishing and got off all the TIM that had been resistant to my previous efforts. The result was a shinier, smoother surface, incidental to ridding the surface of the old TIM. The new application of AS was very thin and my temps look pretty good as well.

 

[m0r7if3r]

Hmm, I am probably being too analytical then. Where is the weak link then? The whole point of TIM is not to improve metal-to-metal conductivity, its to remove air, which is an insulator.

I am a mechanical engineer and I work at a nuclear power plant, so all of this heat transfer business is pretty near and dear to my heart. On our steam generators, we actually do a scrubbing process to increase the amounts of surface roughness because when the plates are trued up they are too "finely finished" (yes, they are about as flat as you can get, laser defined leveling process, CAD controlled, pretty damn accurate haha), and polishing the surface would reduce the performance of the steam generators (more than you would expect!).


Devgoat -> I appreciate that bit, however, did you use the same TIM? Did you apply the same amount in the same way? Same ambient temps? Thats just not a proper "test" to define this honestly.

Now if there are actual untrue grooves and markings/warpings in the plates, then absolutely this is necessary, I've just never seen a plate or CPU that seemed to "need" polishing. If thats the case, then it makes a lot more sense to me. Getting an even surface is far more important than any "polishing". I'm just seeing a lot of "polished" for polished sakes out there, and it just seems really strange to me. Simple polishing will not true up anything, and sometimes will make things worse if you use a silicone based compound. Thats more of what Im speaking of.

I just think of it as metal on metal > metal on tim on metal > metal on air on metal. If you're not using a block to make sure the surface is flat, then you're probably just wasting your time, but I'm unfamilar with the process, so I'll decline to comment.

I presume in your nuclear power example that there are two factors at work. I'd bet that it's not multi-media (which is to say, it's something like a WC radiator, where it's metal on air, and nothing else), so the increased surface area beats better mating. Secondly, it wouldn't surprise me if the surface being scuffed was having oxidation removed...but not know the spec I can only guess. Feel free to tell me I'm wrong, my expertise is not mechanical.

 

[bob4933]

I just think of it as metal on metal > metal on tim on metal > metal on air on metal. If you're not using a block to make sure the surface is flat, then you're probably just wasting your time, but I'm unfamilar with the process, so I'll decline to comment.

I presume in your nuclear power example that there are two factors at work. I'd bet that it's not multi-media (which is to say, it's something like a WC radiator, where it's metal on air, and nothing else), so the increased surface area beats better mating. Secondly, it wouldn't surprise me if the surface being scuffed was having oxidation removed...but not know the spec I can only guess. Feel free to tell me I'm wrong, my expertise is not mechanical.

Metal on metal, conduction heat from water to another water system. Dual water system using plates to transfer heat from one pipe to another basically. Completely the same concept. Oxidation removal and "truing the surface" are completely legitimate reasons for this. I never expected that a "good quality" cpu heatsink would be warped from the factory, and I guess that is indeed the case!

 

[NiHaoMike]

I am a mechanical engineer and I work at a nuclear power plant, so all of this heat transfer business is pretty near and dear to my heart. On our steam generators, we actually do a scrubbing process to increase the amounts of surface roughness because when the plates are trued up they are too "finely finished" (yes, they are about as flat as you can get, laser defined leveling process, CAD controlled, pretty damn accurate haha), and polishing the surface would reduce the performance of the steam generators (more than you would expect!).

They do something similar in HVAC as well. Look at any air conditioning system and you'll note that the condenser and evaporator coil fins are made from textured aluminum sheets instead of flat aluminum sheets. That makes the heat exchanger more restrictive but the improvement in thermal transfer is worth it. The refrigerant side of air coils are rarely textured as the poor conductivity of air is the limiting factor, but on water cooled condensers, the tubes are often twisted and textured on both sides in order to increase effective surface area.

But if you're trying to transfer heat between two solids of high thermal conductivity (as opposed to between a solid and a liquid or gas), the ideal is to have the two solids in direct contact with no gap in between. Since that's not possible, the best we could do is to get the surfaces as flat as possible and fill in remaining gaps with thermal grease.

 

[MongGrel]

Unless a cooler is really out of flatness I guess is kinda irrelevant these days, I lapped my TRUE copper years ago.

It probably wasn't necessary on that even at the time, it was pretty flat to begin with.

The new Noctua I didn't even bother lapping, it looked that flat, most manufacturers have figure that one out themselves it looks.

It's probably redundant these days.

I suppose you have to look at what arrives perhaps.

I doubt making a cooler contact a CPU like wringing Jo Blocks without TIM would make that large a difference in reality depending on several other factors.

http://en.wikipedia.org/wiki/Johanssen_blocks